feat: add local structured substitute benchmark strategy

[binaryaaron]

Summary

Adds a benchmark-only local structured substitute path that builds deterministic replacement maps for supported structured labels. This lets the benchmark suite compare DataDesigner substitute output against a local, non-LLM substitute baseline without changing Anonymizer defaults.

Stack

• Base: feat: anonymizer measurement instrumentation and benchmark tooling #177 measurement instrumentation and benchmark infrastructure
• This PR: local structured substitute replacement-map strategy
• Next: feat: add regex-backed detection benchmark strategies #183 regex-backed detection benchmark strategies

What changed

• Add structured_substitute.py for deterministic replacements on supported structured labels.
• Add benchmark strategy wiring and replay tooling for structured substitute comparisons.
• Add focused tests for replacement-map behavior, replay output, benchmark wiring, and analysis fields.
• Document the strategy as a benchmark probe, not a public product default.

Validation

• uv run --frozen ruff format ...
• uv run --frozen ruff check ...
• uv run pytest tests/engine/test_structured_substitute.py tests/test_measurement.py tests/tools/test_measurement_tools.py tests/tools/test_replacement_strategies.py tests/tools/test_replay_replacement_strategies.py tests/tools/test_benchmark_output_analysis.py tests/tools/test_compare_strategy_pairs.py tests/tools/test_screen_strategy_comparisons.py -q
• git diff --check

Focused suite result: 150 passed, with existing DataDesigner provider deprecation warnings.